Safety monitoring method, system and equipment for solid hydrogen storage test bed and medium

Abstract

The invention provides a safety monitoring method, system and equipment for a solid hydrogen storage test bed and a medium. The method comprises the following steps: collecting real-time information; the real-time information comprises real-time sensor data and a real-time video stream; performing time sequence alignment processing on the real-time information to obtain current information; the current information comprises current sensor data and a current video stream; judging whether the current sensor data exceeds a dynamic threshold value or not, and predicting a subsequent data trend corresponding to the current sensor data and performing visual analysis on a current video stream by using AI; when the current sensor data exceeds a dynamic threshold value, an alarm is given; under the condition that the predicted follow-up data trend does not meet the dynamic condition, early warning is carried out; and under the condition that an analysis result obtained by the visual analysis indicates hydrogen leakage, determining a leakage position and closing an electromagnetic valve associated with the leakage position. According to the scheme, multi-mode signal fusion, rapid and comprehensive leakage condition identification, adaptive threshold alarm and protocol adaptive compatibility can be realized.

Description

Technical Field

[0001] This invention relates to the field of hydrogen energy safety technology, and in particular to a safety monitoring method, system, equipment and medium for a solid hydrogen storage test bench. Background Technology

[0002] Solid-state hydrogen storage test benches are core and critical equipment in the hydrogen energy research and development system. They are mainly used for performance evaluation of hydrogen storage materials, structural optimization of hydrogen storage systems, and verification of related processes. Their operational stability and monitoring accuracy directly determine the reliability of research and development data and the safety of the testing process.

[0003] The operating environment of a solid-state hydrogen storage test rig is significantly complex and highly hazardous: First, the operating pressure covers a high-pressure range of 0.1~1.0MPa, with core pipelines and sealing components subjected to high-pressure loads for extended periods. Second, the temperature range is extremely wide, requiring adaptation to a broad temperature range of -40~80℃, with extreme temperatures easily accelerating material aging and seal failure. Third, the test system extensively uses hydrogen-embrittlement-sensitive materials (such as some high-strength steels and aluminum alloys), which are prone to hydrogen embrittlement damage and structural failure under the combined effects of a hydrogen atmosphere and cyclic loads. Due to the combined influence of these operating conditions and material properties, during long-term cyclic testing or R&D experiments, the solid-state hydrogen storage test rig is highly susceptible to failures such as material fatigue cracking, valve seal failure, and loose pipeline joints, which can induce hydrogen leakage, abnormal system pressure, and other safety hazards, seriously threatening the personal safety of test personnel and the property safety of test equipment.

[0004] Currently, the industry's operational monitoring of solid-state hydrogen storage test benches still relies mainly on traditional methods. These monitoring methods primarily depend on a single type of sensor (such as mechanical pressure gauges and conventional temperature sensors) and manual inspections, which cannot adapt to the complex and high-risk operating conditions mentioned above and have significant systemic defects, as specifically manifested as follows: (1) Severe lag in leak detection: Existing leak detection methods mostly rely on catalytic combustion hydrogen concentration sensors, whose response time generally exceeds 10 seconds, and require the hydrogen concentration to accumulate to 1% LEL (approximately 400 ppm) or higher before triggering an alarm. Due to the characteristics of hydrogen's rapid diffusion and wide combustion and explosion range, by the time the sensor triggers an alarm, the hydrogen leak has already lasted for tens of seconds, and the leak amount has accumulated to a dangerous level, which can easily lead to a combustion and explosion accident, and cannot meet the real-time safety protection requirements of the test bench.

[0005] (2) Fragmented multi-source data and lack of identification of coupling risks: The core monitoring parameters of the test bench, such as pressure, temperature and flow rate, are acquired through independent acquisition modules. The modules lack a unified time sequence alignment and data fusion mechanism, which makes it impossible to effectively explore the correlation characteristics between parameters. For example, when coupled faults such as "normal temperature but sudden pressure drop" or "stable flow rate but abnormal temperature rise" occur, the existing monitoring system cannot identify such potential risks and is prone to misjudgment or omission of faults due to the judgment deviation of a single parameter.

[0006] (3) Visual inspection has inherent blind spots: According to the national standard "Basic Requirements for Hydrogen System Safety" (GB / T29729-2013), solid hydrogen storage test benches need to be visually inspected regularly. Currently, the industry mainly relies on manual observation of the color change of hydrogen detection tape or the frost on the pipeline surface to make an initial judgment of leakage. This method is highly subjective and affected by factors such as the staff's sense of responsibility, inspection frequency, and ambient light, resulting in a very high rate of missed detection. Industry survey data shows that the missed detection rate of this type of manual visual inspection exceeds 30%, and it is particularly unable to identify early leakage characteristics such as slight discoloration of tape and slight frost.

[0007] (4) Closed communication protocols and extremely poor scalability: Since the instruments and equipment on the test bench are mostly purchased from different manufacturers (such as Emerson pressure transmitters, Siemens temperature controllers, Honeywell flow meters, etc.), the communication protocols used by each device are significantly different, covering multiple protocol types such as Modbus RTU / TCP, HART, and Profibus. The existing SCADA (data acquisition and monitoring system) requires manual configuration of register mapping to achieve multi-device access. When R&D requirements change and new sensors or instruments need to be added or replaced, the system needs to be shut down for 2 to 4 hours for protocol debugging and system adaptation, which seriously affects R&D efficiency and cannot meet the testing requirements of high-frequency iteration in the R&D stage.

[0008] To address the shortcomings of traditional monitoring methods, the industry has undertaken some technological improvements in an attempt to enhance the real-time performance and reliability of solid-state hydrogen storage test benches. However, each improvement scheme still has its limitations and has failed to fundamentally solve the existing problems, as detailed below: Option 1: Connect a hydrogen concentration sensor in series at the outlet of the cylinder valve and set the alarm threshold to 500 ppm to improve response speed. However, this improvement has three major problems: ① The monitoring dimension is limited, only monitoring the concentration at a single point at the outlet of the cylinder valve, which cannot accurately locate the leak source and is not conducive to troubleshooting; ② It does not integrate the changing trends of related parameters such as pressure and temperature. When the ambient humidity exceeds 85%RH, the sensor is susceptible to water vapor interference, resulting in a false alarm rate as high as 22%; ③ It lacks a visual signal verification mechanism, and cannot effectively identify early leak characteristics such as slight discoloration of the tape, failing to fill the blind spots of visual inspection.

[0009] Option 2: Establish a two-dimensional pressure-temperature (PT) correlation model. An early warning is triggered when the ratio of the pressure change rate to the temperature change rate (ΔP / ΔT) exceeds an empirical threshold. While this option achieves correlation analysis of some parameters, it still has significant shortcomings: ① It does not incorporate visual monitoring signals, making it insensitive to minute leaks such as valve micro-leakage with a pressure drop of less than 0.01 MPa, resulting in a high false alarm rate; ② The early warning threshold is a fixed empirical value, unable to adapt to seasonal temperature drift, with a significantly increased false alarm rate of 18% in low-temperature winter environments; ③ The cloud-based centralized data processing architecture requires data to be uploaded to the cloud for analysis and calculation, resulting in a 3-5 second transmission and processing delay, which cannot meet real-time monitoring requirements.

[0010] Option 3: Employing a TDLAS (Tunable Diode Laser Absorption Spectroscopy) laser spectral sensor to improve hydrogen detection sensitivity to 10 ppm. While this sensor offers significant advantages in detection sensitivity, its practicality is limited: ① The equipment is expensive, with a single-point deployment cost exceeding 30,000 RMB. Due to R&D budget constraints, it cannot achieve full coverage monitoring across multiple test sites; ② The laser detection window is easily obstructed by silicone tape or environmental dust used during testing, requiring frequent shutdowns for maintenance and affecting test continuity; ③ It only outputs hydrogen concentration detection values ​​and lacks a linkage control mechanism with the existing SCADA system, failing to automatically trigger emergency actions such as solenoid valve shutdown, resulting in an incomplete protection loop.

[0011] Option 4: Use an infrared camera to identify the "cold spot" characteristics caused by cryogenic hydrogen leaks. This option attempts to fill the gap in visual monitoring, but it has several technical bottlenecks: ① Due to the limitation of camera resolution, it cannot effectively identify trace leaks with a leakage rate of less than 0.5 g / s, resulting in insufficient monitoring sensitivity; ② It is easily affected by ambient light interference, with a false detection rate as high as 40% in daylight under direct sunlight or reflected light; ③ It only identifies the "cold spot" characteristics of cryogenic leaks, failing to solve the problem of early visual feature extraction and identification of hydrogen detection tape color changes, resulting in incomplete coverage of visual monitoring scenarios.

[0012] In summary, existing monitoring technologies for solid-state hydrogen storage test benches, whether traditional monitoring methods or subsequent improvement schemes, have failed to overcome three core common defects: First, the limitation of single-modal perception, relying only on a single type of sensor signal (concentration, pressure, temperature, or vision), without achieving deep fusion of multi-modal signals; second, the problem of static thresholds, with warning thresholds mostly being fixed empirical values, unable to adapt to fluctuations in operating conditions such as ambient temperature drift and humidity changes; and third, the defect of protocol closure, with high cost and poor scalability for protocol compatibility of instruments from multiple manufacturers, unable to meet the high-frequency iteration needs of the R&D stage.

[0013] Therefore, existing monitoring technologies are insufficient to meet the complex operating conditions of solid-state hydrogen storage test benches, which involve high pressure, wide temperature range, and mixed connections of instruments from multiple manufacturers. In particular, they cannot simultaneously address the core requirements of real-time monitoring, accuracy, and scalability during the research and development process. Developing a monitoring system for solid-state hydrogen storage test benches that can achieve multi-modal signal fusion, rapid and comprehensive identification of leakage conditions, precise location of leakage sources, adaptive condition early warning, adaptive threshold alarm, and adaptive protocol compatibility has become an urgent technical challenge to be solved in the current hydrogen energy research and development field. Summary of the Invention

[0014] To overcome the shortcomings of the existing technology, this invention provides a safety monitoring method, system, equipment, and medium for solid-state hydrogen storage test benches. It can achieve multi-modal signal fusion, rapid and comprehensive identification of leakage, precise location of leakage sources, adaptive condition early warning, adaptive threshold alarm, and protocol adaptive compatibility, thereby solving the three core problems of insufficient real-time performance, difficulty in multi-parameter coordination, and delayed leakage early warning in existing solid-state hydrogen storage test bench monitoring.

[0015] To solve the above-mentioned technical problems, the present invention provides the following technical solution: According to a first aspect of the present invention, a safety monitoring method for a solid-state hydrogen storage test bench is provided, comprising: Collect real-time information; the real-time information includes real-time sensor data and real-time video stream; The real-time information is time-aligned to obtain the current information; the current information includes the current sensor data and the current video stream. Determine whether the current sensor data exceeds a dynamic threshold, and use AI to predict the subsequent data trend corresponding to the current sensor data and perform visual analysis on the current video stream; An alarm is triggered if the current sensor data exceeds the dynamic threshold; a warning is issued if the predicted trend of subsequent data does not meet the dynamic conditions; and the location of the leak is determined and the solenoid valve associated with the leak location is closed if the analysis results obtained from visual analysis indicate a hydrogen leak.

[0016] In one exemplary implementation, the dynamic threshold includes a pressure safety threshold and a temperature safety threshold; The pressure safety threshold includes a pressure upper limit threshold and a pressure fluctuation rate threshold; The formula for calculating the upper limit of pressure is as follows: ; Where, Peq(Tenv): the hydrogen absorption equilibrium pressure at the current ambient temperature; ksafe: Safety factor for hydrogen storage materials; Pcontainer: Rated pressure resistance of hydrogen storage container; kcontainer: Safety factor for hydrogen storage containers; RHmax: Upper limit of humidity for hydrogen storage materials; The pressure fluctuation rate threshold is determined based on ambient temperature fluctuation and ambient humidity. The temperature safety threshold includes an upper temperature threshold, a lower temperature threshold, and a temperature change rate threshold. The formula for calculating the upper temperature threshold is as follows: ; ΔTmax: The maximum permissible reaction temperature rise, determined based on the ambient humidity. Tmat_decompose: Decomposition temperature of hydrogen storage material; The formula for calculating the lower temperature threshold is as follows: ; ΔTmin: Maximum allowable reaction temperature drop, set to a fixed value; Tmat_brittle: Embrittlement temperature of hydrogen storage material; The temperature change rate threshold is determined based on the hydrogen storage stage, which includes a hydrogen absorption stage and a hydrogen release stage; when in the hydrogen absorption stage, the temperature change rate threshold is determined based on the ambient humidity.

[0017] In one exemplary implementation, the Peq(Tenv) is obtained based on a PCT curve query or calculated using the Van'tHoff equation; The PCT curve is obtained based on historical data statistics; The formula for calculating the Van'tHoff equation is as follows: ; ΔH: Enthalpy change of hydrogen absorption in hydrogen storage materials; ΔS: Entropy change of hydrogen absorption in hydrogen storage material; R: gas constant (8.314 J / L) ); T: Ambient temperature (unit: K, T=Tenv+273.15).

[0018] In one exemplary implementation, the step of using AI to predict the subsequent data trend corresponding to the current sensor data specifically includes: The target LSTM prediction model is used to predict the subsequent data trend corresponding to the current sensor data; the subsequent data trend is the target pressure-temperature curve within a preset future time.

[0019] In one exemplary embodiment, the method further includes a training process for the target LSTM prediction model, specifically comprising: Acquire historical data and uniformly divide the time series windows; The pressure, temperature, ambient temperature, and ambient humidity rate of change features are extracted from the historical data and normalized to obtain the source domain data. Construct an initial LSTM prediction model and output the initial pressure-temperature curve for a preset future time based on the source domain data; Based on the initial pressure-temperature curve within a preset future time and the actual pressure-temperature curve within a preset future time obtained from the source domain data, the parameters of the initial LSTM prediction model are adjusted to obtain the current LSTM prediction model. The source domain data is input into the current LSTM prediction model, and the current pressure-temperature curve is output within a preset future time. Determine the deviation between the current pressure-temperature curve and the actual pressure-temperature curve within a preset future time period; If the deviation does not meet the deviation condition, the parameters of the current LSTM prediction model are adjusted, and the steps of inputting the source domain data into the current LSTM prediction model and judging the deviation between the current pressure-temperature curve and the actual pressure-temperature curve within a preset future time are repeated until the deviation meets the deviation condition. Then, the current LSTM prediction model is output as the target LSTM prediction model.

[0020] In one exemplary implementation, the step of using AI to perform visual analysis on the current video stream specifically includes: Visual analysis of the current video stream is performed using target visual models; the target visual models include a target color recognition model, a target YOLOv5 model, and a target ResNet18 classification model.

[0021] In one exemplary embodiment, the method further includes: Automatic polling extension interface; If a new sensor is detected to be connected to the expansion interface, an identification command is sent to the new sensor. Obtain the identity information sent by the newly added sensor; Automatic identification is achieved by comparing the identity information with the built-in compatibility library; Based on the identified communication protocol type, the corresponding parsing library is automatically loaded; Negotiate and lock the optimal communication parameters with the newly added sensor; The system parses the raw data format of the newly added sensor, performs normalization transformation and anomaly filtering, and outputs standardized data that matches the test bench. Based on the measurement parameter type of the newly added sensor, a safety threshold template or a user-matched threshold is automatically assigned.

[0022] According to a second aspect of the present invention, a safety monitoring system for a solid-state hydrogen storage test bench is provided, implemented using any of the safety monitoring methods for solid-state hydrogen storage test benches described above, the system comprising: The information acquisition module is used to acquire real-time information, including real-time sensor data and real-time video streams. The information preprocessing module is used to perform time-series alignment processing on the real-time information to obtain the current information; the current information includes the current sensor data and the current video stream. The real-time monitoring module is used to determine whether the current sensor data exceeds a dynamic threshold, and to use AI to predict the subsequent data trend corresponding to the current sensor data and to perform visual analysis on the current video stream. The safety rule triggering module is used to issue an alarm when the current sensor data exceeds the dynamic threshold; to issue a warning when the predicted subsequent data trend does not meet the dynamic conditions; and to determine the leak location and close the solenoid valve associated with the leak location when the analysis results obtained from visual analysis indicate a hydrogen leak.

[0023] According to a third aspect of the present invention, an electronic device is provided, including a processor and a memory, wherein the memory stores at least one instruction or at least one program, the at least one instruction or the at least one program being loaded and executed by the processor to implement any of the above-described safety monitoring methods for a solid-state hydrogen storage test bench.

[0024] According to a fourth aspect of the present invention, a computer-readable storage medium is provided, wherein the storage medium stores at least one instruction or at least one program, wherein the at least one instruction or the at least one program is loaded and executed by a processor to implement any of the above-described safety monitoring methods for a solid-state hydrogen storage test bench.

[0025] By adopting the above technical solution, the present invention has the following beneficial effects: (1) This invention integrates three core monitoring modes: real-time monitoring of sensor data, AI multi-data fusion prediction, and AI visual recognition real-time monitoring, to construct a collaborative safety monitoring system with a full-chain "early warning-alarm-processing" mechanism. Compared with existing single monitoring methods, multi-modal signal fusion can achieve comprehensive and three-dimensional perception of the test bench's operating status. Through in-depth mining and fusion analysis of multi-source data using AI algorithms, potential safety risks can be predicted in advance, avoiding missed judgments caused by single data deviations. At the same time, the coordinated linkage of early warning, alarm, and processing can respond quickly in the early stages of risks and take intervention measures before risks escalate, significantly reducing the probability of safety accidents and improving the safety and stability of the test bench's operation.

[0026] (2) To address the problem that early signs of hydrogen leakage in solid-state hydrogen storage tests are not obvious and are easily overlooked, this invention employs AI visual recognition technology to accurately capture key early signs of leakage, such as changes in the color of hydrogen conveyor belts, pipe deformation, and smoky gas flow. This design overcomes the limitations of traditional leak monitoring, which relies on a single physical parameter and is difficult to identify early minor leaks. It can quickly and comprehensively identify leak signs in the early stages of a leak, allowing sufficient time for subsequent early warning responses and emergency handling, curbing the expansion of leak risks from the source, and significantly reducing casualties and property losses caused by leak accidents.

[0027] (3) This invention achieves precise location of the leak source through AI visual recognition technology. Compared with existing methods that rely on sensor arrays for rough location or manual investigation, it has the advantages of high positioning accuracy and fast response speed. Precise leak source location can directly provide clear guidance for emergency response, avoid blind operation during the response process, help staff quickly locate risk points and take targeted sealing, isolation and other measures, effectively shorten emergency response time, and reduce the spread and severity of leak accidents.

[0028] (4) This invention innovatively adopts an adaptive threshold early warning and alarm mechanism, wherein the early warning slope is designed with a non-fixed value, and the alarm threshold can be dynamically adjusted according to environmental conditions. This design overcomes the shortcomings of traditional fixed threshold monitoring methods that are difficult to adapt to changes in environmental parameters such as temperature, humidity, and air pressure around the test bench. It can match the dynamic changes in environmental conditions in real time, accurately distinguish between normal fluctuations and abnormal risks, significantly improve the accuracy of early warning and alarm, effectively reduce the risk of false alarm interference and missed alarms caused by threshold rigidity, and ensure the effectiveness and practicality of the monitoring system.

[0029] (5) This invention has protocol adaptive compatibility, and can be adapted to sensors, visual acquisition devices and test bench control units of different types and communication protocols. Compared with the limitations of existing monitoring systems with poor compatibility and only able to adapt to specific models of equipment, this design greatly improves the versatility and flexibility of the system. It can be directly applied to solid hydrogen storage test benches of different specifications and configurations without the need for large-scale modification of existing test bench equipment, which significantly reduces the deployment cost and promotion difficulty of the system and is conducive to the large-scale application of the technology.

[0030] (6) This invention achieves automated and intelligent upgrades to safety monitoring through the deep integration of AI technology and multimodal monitoring, reducing reliance on manual monitoring and intervention. On the one hand, it reduces subjective errors and labor intensity in the manual monitoring process; on the other hand, through early warning, precise positioning and rapid response, it reduces equipment damage and maintenance costs caused by safety accidents, while extending the service life of the core equipment of the test bench, comprehensively improving the operation and maintenance management efficiency of the solid hydrogen storage test bench, and reducing the overall operation and maintenance cost. Attached Figure Description

[0031] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0032] Figure 1 A flowchart illustrating a safety monitoring method for a solid-state hydrogen storage test bench provided in an embodiment of the present invention; Figure 2 A structural block diagram of a safety monitoring system for a solid-state hydrogen storage test bench provided in an embodiment of the present invention; Figure 3 This is a hardware structure block diagram of an electronic device for running a safety monitoring method for a solid-state hydrogen storage test bench, provided in an embodiment of the present invention. Detailed Implementation

[0033] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0034] The term "an embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the invention. In the description of the embodiments of the invention, it should be understood that the terms "upper," "lower," "top," "bottom," etc., indicating orientation or positional relationships based on the orientation or positional relationships shown in the accompanying drawings, are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the invention. Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined with "first" and "second" may explicitly or implicitly include one or more of that feature. Moreover, the terms "first," "second," etc., are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein.

[0035] Please see Figure 1 The diagram shows a flowchart of a safety monitoring method for a solid-state hydrogen storage test bench provided by an embodiment of the present invention. This safety monitoring method for a solid-state hydrogen storage test bench includes the following steps: Step S1: Acquire real-time information; real-time information includes real-time sensor data and real-time video stream; Step S2: Perform time-series alignment processing on the real-time information to obtain the current information; the current information includes the current sensor data and the current video stream; Step S3: Determine whether the current sensor data exceeds the dynamic threshold, and use AI to predict the subsequent data trend corresponding to the current sensor data and perform visual analysis on the current video stream; Step S4: If the current sensor data exceeds the dynamic threshold, issue an alarm; if the predicted subsequent data trend does not meet the dynamic conditions, issue a warning; if the analysis results obtained from visual analysis indicate a hydrogen leak, determine the leak location and close the solenoid valve associated with the leak location.

[0036] In an optional embodiment, the dynamic threshold in step S3 above includes a pressure safety threshold and a temperature safety threshold; Pressure safety thresholds include upper pressure limit thresholds and pressure fluctuation rate thresholds; The formula for calculating the upper pressure threshold is as follows: ; Where, Peq(Tenv): the hydrogen absorption equilibrium pressure at the current ambient temperature; ksafe: Safety factor for hydrogen storage materials; Pcontainer: Rated pressure resistance of hydrogen storage container; kcontainer: Safety factor for hydrogen storage containers; RHmax: Upper limit of humidity for hydrogen storage materials; The pressure fluctuation rate threshold is determined based on ambient temperature fluctuations and ambient humidity. Temperature safety thresholds include upper temperature threshold, lower temperature threshold, and temperature change rate threshold; The formula for calculating the upper temperature threshold is as follows: ; ΔTmax: Maximum permissible reaction temperature rise, determined based on ambient humidity; Tmat_decompose: Decomposition temperature of hydrogen storage material; The formula for calculating the lower temperature threshold is as follows: ; ΔTmin: Maximum allowable reaction temperature drop, set to a fixed value; Tmat_brittle: Embrittlement temperature of hydrogen storage material; The temperature change rate threshold is determined based on the hydrogen storage stage, which includes the hydrogen absorption stage and the hydrogen release stage; when in the hydrogen absorption stage, the temperature change rate threshold is determined based on the ambient humidity.

[0037] In an optional embodiment, the above-mentioned Peq(Tenv) is obtained by querying the PCT curve (pressure-composition-temperature) or by calculating it through the Van'tHoff equation; The PCT curve is obtained based on historical data statistics; The formula for calculating the Van't Hoff equation is as follows: ; ΔH: Enthalpy change of hydrogen absorption in hydrogen storage materials; ΔS: Entropy change of hydrogen absorption in hydrogen storage material; R: gas constant (8.314 J / L) ); T: Ambient temperature (unit: K, T=Tenv+273.15).

[0038] In an optional embodiment, step S3 above, which uses AI to predict the subsequent data trend corresponding to the current sensor data, specifically includes: The target LSTM prediction model is used to predict the subsequent data trend corresponding to the current sensor data; the subsequent data trend is the target pressure-temperature curve within a preset future time.

[0039] In an optional embodiment, the above method further includes a training process for the target LSTM prediction model, specifically including: Acquire historical data and uniformly divide the time series windows; Extract the characteristics of pressure, temperature, ambient temperature, and ambient humidity change rate from historical data, and perform normalization processing to obtain source domain data. Construct an initial LSTM prediction model and output the initial pressure-temperature curve for a preset future time based on the source domain data; Based on the initial pressure-temperature curve within a preset future time and the actual pressure-temperature curve within a preset future time obtained from the source domain data, the parameters of the initial LSTM prediction model are adjusted to obtain the current LSTM prediction model. Input the source domain data into the current LSTM prediction model and output the current pressure-temperature curve within a preset future time. Determine the deviation between the current pressure-temperature curve and the actual pressure-temperature curve within a preset future timeframe; If the deviation does not meet the deviation condition, the parameters of the current LSTM prediction model are adjusted, and the source domain data is repeatedly input into the current LSTM prediction model until the deviation between the current pressure-temperature curve and the actual pressure-temperature curve within the preset future time is judged. This process continues until the deviation meets the deviation condition, at which point the current LSTM prediction model is output as the target LSTM prediction model.

[0040] In an optional embodiment, step S3 above, which involves using AI to perform visual analysis of the current video stream, specifically includes: Visual analysis of the current video stream is performed using target visual models, which include a target color recognition model, a target YOLOv5 model, and a target ResNet18 classification model.

[0041] In an optional embodiment, the above method further includes a training process for the target color recognition model, specifically including: Acquire historical image frame data of hydrogen tape in different color states; Extract color space features, color histograms, and color moment features from historical image frame data, and perform normalization processing to obtain sample color data; An initial color recognition model was constructed based on a convolutional neural network architecture, and the initial hydrogen tape color classification results were output based on sample color data. Based on the initial hydrogen tape color classification results and the actual hydrogen tape color classification results obtained from the sample color data annotation, the parameters of the initial color recognition model are adjusted to obtain the current color recognition model. Input the color training set into the current color recognition model and output the current hydrogen tape color classification result; Determine the discrepancy between the current hydrogen tape color classification result and the actual hydrogen tape color classification result; If the deviation does not meet the deviation condition, the parameters of the current color recognition model are adjusted, and the sample color data is repeatedly input into the current color recognition model until the deviation between the current hydrogen tape color classification result and the actual hydrogen tape color classification result is judged. This process continues until the deviation meets the deviation condition, at which point the current color recognition model is output as the target color recognition model.

[0042] In an optional embodiment, the above method further includes a training process for the target YOLOv5 model, specifically including: Acquire historical image or video frame data of the pipeline under different deformation states; Extract image texture features, edge features, and grayscale features from historical image or video frame data, and perform pixel value normalization and image enhancement preprocessing to obtain sample pipeline images; Construct an initial YOLOv5 model and output initial pipe deformation detection results based on preprocessed pipe images; Based on the initial pipeline deformation detection results and the actual pipeline deformation detection results obtained from the annotation of sample pipeline images, the parameters of the initial YOLOv5 model are adjusted to obtain the current YOLOv5 model. Input the sample pipe image into the current YOLOv5 model and output the current pipe deformation detection result; Determine the deviation between the current pipeline deformation detection results and the actual pipeline deformation detection results; If the deviation does not meet the deviation condition, the parameters of the current YOLOv5 model are adjusted, and the sample pipeline image is repeatedly input into the current YOLOv5 model until the deviation between the current pipeline deformation detection result and the actual pipeline deformation detection result is judged. This process continues until the deviation meets the deviation condition, at which point the current YOLOv5 model is output as the target YOLOv5 model.

[0043] In an optional embodiment, the above method further includes a training process for the target ResNet18 classification model, specifically including: Acquire historical image or video frame data showing the presence or absence of smoke-like airflow; Extract image spatial features, texture features, and gray-level co-occurrence matrix features from historical image or video frame data, and perform image standardization and size uniformity preprocessing to obtain sample airflow images; An initial ResNet18 classification model is constructed, and the initial classification results of the smoke-like airflow are output based on the sample airflow images; Based on the initial classification results of the smoke-like airflow and the actual classification results of the smoke-like airflow obtained from the source domain data annotation, the parameters of the initial ResNet18 classification model are adjusted to obtain the current ResNet18 classification model. Input the sample airflow image into the current ResNet18 classification model, and output the current smoke-like airflow classification result; Determine the discrepancy between the current classification result of the smoke-like airflow and the actual classification result of the smoke-like airflow; If the deviation does not meet the deviation condition, the parameters of the current ResNet18 classification model are adjusted, and the sample airflow image is repeatedly input into the current ResNet18 classification model until the deviation between the current smoke-like airflow classification result and the actual smoke-like airflow classification result is judged. This process continues until the deviation meets the deviation condition, at which point the current ResNet18 classification model is output as the target ResNet18 classification model.

[0044] In an optional embodiment, the dynamic condition in step S4 above is the pressure-temperature curve warning slope under different ambient temperatures and humidity levels determined based on historical data.

[0045] In an optional embodiment, the method further includes a step of supporting plug-and-play functionality for new sensors, specifically including: Automatic polling extension interface; When a new sensor is detected to be connected to the expansion interface, a recognition command is sent to the new sensor. Obtain the identity information sent by the newly added sensor; Automatic identification is achieved by comparing identity information with the built-in compatible library; Based on the identified communication protocol type, the corresponding parsing library is automatically loaded; Negotiate and lock in the optimal communication parameters with the new sensor; The system parses the raw data format of the newly added sensors, performs normalization transformation and anomaly filtering, and outputs standardized data that matches the test bench. Based on the measurement parameter type of the newly added sensor, automatically assign a safety threshold template or a user-matched threshold.

[0046] In the first practical application scenario, this safety monitoring method for solid-state hydrogen storage test benches is applied to small-scale solid-state hydrogen storage test benches. The test benches operate at pressures ranging from 0.2 to 0.8 MPa and at temperatures ranging from -20 to 60°C. The system deployment is as follows: 1. Hardware configuration: (1) Sensor layer: Install Emerson explosion-proof pressure transmitter (range 0~1.0MPa), PT100 thermal resistance (-50~150℃), and Siemns flow meter (DN25), and connect to the industrial control computer via RS485 bus.

[0047] (2) Visual layer: Deploy Hikvision 1080P infrared cameras to focus on the pipe weld seam and hydrogen indicator tape area, and use NVIDIA Jetson Xavier AI box for real-time image analysis.

[0048] (3) Control layer: Siemens IPC677E industrial computer is used, which integrates RS485 / HART protocol parsing module and outputs control signals to solenoid valves and audible and visual alarms.

[0049] 2. Software Configuration: (1) SCADA platform: Based on Siemens WinCC, it realizes real-time visualization of pressure, temperature and flow data, historical trend recording and alarm management.

[0050] (2) AI module: ResNet18 classification model is used to identify early leak characteristics; LSTM model predicts the trend of the next 5 minutes based on the pressure sequence of the previous 10 minutes and dynamically adjusts the alarm threshold; YOLOv5s model is used to train pipeline deformation detection.

[0051] 3. Workflow: Data acquisition: The sensor uploads data to the industrial computer every 1 second. The Modbus protocol automatically enumerates newly added instruments, and the scanning time is less than 30 seconds.

[0052] AI visual analysis: The camera captures an image every 0.5 seconds, and the AI ​​model calculates the changes in the RGB values ​​of the tape, detects pipe deformation, and identifies airflow conditions.

[0053] Early warning-alarm-processing linkage: (1) Alarm: An audible and visual alarm is triggered when the pressure is >0.75MPa; (2) Warning: The LSTM detects an abnormal slope of the pressure curve (such as a drop rate of >0.02MPa / min within 10 minutes) and issues a warning 5 minutes in advance; (3) Handling: When the AI ​​identifies that the tape color turns pink, the pipe is deformed, or a smoky airflow appears, the system automatically closes the upstream solenoid valve and records the location of the leak.

[0054] 4. Effect Verification: Safety Enhancement: During the 3-month trial operation, the system successfully identified 2 micro-leakages (leakage rate <0.1g / s), the response time was shortened from the traditional >10 seconds to 1.5 seconds, and the leak detection rate was improved from 70% to 96%.

[0055] Efficiency optimization: The frequency of manual inspections has been reduced from once per hour to once per day, saving approximately 40% in labor costs.

[0056] Compatibility: When a new Honeywell pressure sensor is added, the system automatically recognizes and integrates it without downtime for debugging, and the deployment time is only 15 minutes.

[0057] In the second practical application scenario, this safety monitoring method for solid-state hydrogen storage test benches is applied to a medium-sized solid-state hydrogen storage test platform. This platform contains multiple independent hydrogen storage pipelines (pressure 0.1~1.0MPa), and the instruments are from multiple manufacturers (Emerson, Siemens, Honeywell), using mixed protocols (Modbus RTU / TCP, HART). The system deployment is as follows: 1. Hardware configuration: (1) Sensor layer: Each pipeline is equipped with a pressure transmitter, a thermal resistor and a flow meter, totaling 12 measuring points, which are connected to the edge industrial control computer through an RS485 hub.

[0058] (2) Visual layer: Four 2-megapixel global shutter cameras are used to cover all pipe connection points and tape areas. The AI ​​box is equipped with Jetson Xavier NX and supports parallel analysis of multiple video streams.

[0059] (3 Control layer: Adopts Advantech industrial computer with built-in multi-protocol adaptive module (supports Modbus, HART, Profibus), outputs control signals to distributed solenoid valve group.

[0060] 2. Software Configuration: (1) SCADA system: developed based on the Ignition platform, realizing multi-pipeline topology visualization, data synchronous storage and alarm hierarchical push.

[0061] (2) AI module: ResNet18 classification model is used to identify early leak characteristics; LSTM model predicts the trend of the next 5 minutes based on the pressure sequence of the previous 10 minutes and dynamically adjusts the alarm threshold; YOLOv5s model is used to train pipeline deformation detection.

[0062] 3. Work Process (1) Automatic protocol adaptation: When a new Siemens HART temperature transmitter is added, the system automatically scans the register address and completes the mapping in less than 25 seconds without stopping the system.

[0063] (2) Multi-source data fusion: The SCADA system performs time-series alignment of pressure, temperature and flow data, and performs cross-validation by combining the tape color score output by AI vision.

[0064] (3) Dynamic threshold warning: The LSTM model adaptively adjusts the alarm slope based on the data from the previous 10 minutes of operation to reduce false alarms.

[0065] (4) Linkage control: When the AI ​​vision detects that the tape on a specific pipeline has changed color and the pressure trend is abnormal, the system will only close the solenoid valve of the corresponding pipeline to avoid overall shutdown.

[0066] 4. Effect Verification: Reduced false alarm rate: Compared with the traditional fixed threshold method, the dynamic threshold mechanism reduces the annual false alarm rate fluctuation to <2%.

[0067] Deployment efficiency: The system supports plug-and-play sensors, and the average time to add a new measurement point is 18 minutes, which is 8 times faster than the traditional method.

[0068] Multimodal collaboration: In a simulated leak test, the system accurately located the leak point by combining the judgment of sudden pressure drop and tape discoloration, with a false alarm rate of 0%.

[0069] Corresponding to the safety monitoring method for a solid-state hydrogen storage test bench provided in the above embodiments, this embodiment of the invention also provides a safety monitoring system for a solid-state hydrogen storage test bench. Since the safety monitoring system for a solid-state hydrogen storage test bench provided in this embodiment corresponds to the safety monitoring method for a solid-state hydrogen storage test bench provided in the above embodiments, the implementation methods of the aforementioned safety monitoring method for a solid-state hydrogen storage test bench are also applicable to the safety monitoring system for a solid-state hydrogen storage test bench provided in this embodiment, and will not be described in detail in this embodiment.

[0070] Please see Figure 2 The diagram shown is a structural block diagram of a safety monitoring system for a solid-state hydrogen storage test bench provided by an embodiment of the present invention; the system includes: 01: Information acquisition module, used to collect real-time information; real-time information includes real-time sensor data and real-time video stream; 02: Information preprocessing module, used to perform time-series alignment processing on real-time information to obtain current information; current information includes current sensor data and current video stream; 03: Real-time monitoring module, used to determine whether the current sensor data exceeds the dynamic threshold, and to use AI to predict the subsequent data trend corresponding to the current sensor data and to perform visual analysis on the current video stream; 04: Safety rule triggering module, used to trigger an alarm when the current sensor data exceeds the dynamic threshold; to issue a warning when the predicted subsequent data trend does not meet the dynamic conditions; and to determine the leak location and close the solenoid valve associated with the leak location when the analysis results obtained from visual analysis indicate a hydrogen leak.

[0071] It should be noted that the system provided in the above embodiments is only illustrated by the division of the above functional modules when implementing its functions. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above. In addition, the system and method embodiments provided in the above embodiments belong to the same concept, and the specific implementation process can be found in the method embodiments, which will not be repeated here.

[0072] This invention also provides an electronic device, including a processor and a memory, wherein the memory stores at least one instruction or at least one program, and the processor loads and executes the at least one instruction or at least one program to implement the safety monitoring method for a solid-state hydrogen storage test bench provided in the above method embodiments.

[0073] Memory can be used to store software programs and modules. The processor executes these stored software programs and modules to perform various functional applications and achieve advanced autonomous driving. Memory can primarily include a program storage area and a data storage area. The program storage area can store the operating system, application programs required for functions, etc.; the data storage area can store data created based on device usage, etc. Furthermore, memory can include high-speed random access memory and non-volatile memory, such as at least one disk storage device, flash memory device, or other volatile solid-state storage device. Accordingly, memory can also include a memory controller to provide the processor with access to the memory.

[0074] The method embodiments provided in this invention can be executed in a computer terminal, server or similar computing device, that is, the above-mentioned electronic device may include a computer terminal, server or similar computing device. Figure 3 This is a hardware structure block diagram of an electronic device for running a safety monitoring method for a solid-state hydrogen storage test bench, as provided in an embodiment of the present invention. Figure 3 As shown, the internal structure of this electronic device may include, but is not limited to, a processor, a network interface, and a memory. The processor, network interface, and memory within the electronic device can be connected via a bus or other means, as illustrated in the embodiments of this specification. Figure 3 Taking the example of a connection between China and Israel via a bus.

[0075] The processor (or CPU, Central Processing Unit) is the computing and control core of the electronic device. The network interface may optionally include a standard wired interface or a wireless interface (such as Wi-Fi, mobile communication interface, etc.). Memory is the storage device in the electronic device used to store programs and data. It is understood that the memory here can be a high-speed RAM storage device or a non-volatile memory device, such as at least one disk storage device; optionally, it can also be at least one storage device located remotely from the aforementioned processor. The memory provides storage space containing the operating system of the electronic device, which may include, but is not limited to: Windows (an operating system), Linux (an operating system), Android (a mobile operating system), iOS (a mobile operating system), etc., and this invention does not limit this; furthermore, the storage space also contains one or more instructions suitable for loading and execution by the processor, which may be one or more computer programs (including program code). In the embodiments of this specification, the processor loads and executes one or more instructions stored in the memory to implement the safety monitoring method for a solid-state hydrogen storage test bench provided in the above method embodiments.

[0076] This invention also provides a computer-readable storage medium storing at least one instruction or at least one program, wherein the at least one instruction or at least one program is loaded and executed by a processor to implement the safety monitoring method for a solid-state hydrogen storage test bench provided in the method embodiment.

[0077] Optionally, in this embodiment, the storage medium may include, but is not limited to, various media capable of storing program code, such as USB flash drives, read-only memory (ROM), random access memory (RAM), portable hard drives, magnetic disks, or optical disks.

[0078] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. Furthermore, specific embodiments have been described above. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps described in the claims can be performed in a different order than that shown in the embodiments and still achieve the desired results. Additionally, the processes depicted in the drawings do not necessarily require a specific or sequential order to achieve the desired results. In some embodiments, multi-sample image classification and parallel processing are also possible or may be advantageous.

[0079] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to interchangeably. Each embodiment focuses on describing the differences from other embodiments. In particular, the system embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions in the method embodiments.

[0080] Those skilled in the art will understand that all or part of the steps of the above embodiments can be implemented by hardware or by a program instructing related hardware. The program can be stored in a computer-readable storage medium, such as a read-only memory, a disk, or an optical disk.

[0081] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A safety monitoring method for a solid-state hydrogen storage test bench, characterized in that, include: Collect real-time information; The real-time information includes real-time sensor data and real-time video stream; The real-time information is time-aligned to obtain the current information; The current information includes current sensor data and current video stream; Determine whether the current sensor data exceeds a dynamic threshold, and use AI to predict the subsequent data trend corresponding to the current sensor data and perform visual analysis on the current video stream; An alarm is triggered if the current sensor data exceeds the dynamic threshold; a warning is issued if the predicted trend of subsequent data does not meet the dynamic conditions; and the location of the leak is determined and the solenoid valve associated with the leak location is closed if the analysis results obtained from visual analysis indicate a hydrogen leak.

2. The safety monitoring method for a solid-state hydrogen storage test bench according to claim 1, characterized in that, The dynamic thresholds include pressure safety thresholds and temperature safety thresholds; The pressure safety threshold includes a pressure upper limit threshold and a pressure fluctuation rate threshold; The formula for calculating the upper limit of pressure is as follows: ; Where, Peq(Tenv): the hydrogen absorption equilibrium pressure at the current ambient temperature; ksafe: Safety factor for hydrogen storage materials; Pcontainer: Rated pressure resistance of hydrogen storage container; kcontainer: Safety factor for hydrogen storage containers; RHmax: Upper limit of humidity for hydrogen storage materials; The pressure fluctuation rate threshold is determined based on ambient temperature fluctuation and ambient humidity. The temperature safety threshold includes an upper temperature threshold, a lower temperature threshold, and a temperature change rate threshold. The formula for calculating the upper temperature threshold is as follows: ； ΔTmax: The maximum permissible reaction temperature rise, determined based on the ambient humidity. Tmat_decompose: Decomposition temperature of hydrogen storage material; The formula for calculating the lower temperature threshold is as follows: ； ΔTmin: Maximum allowable reaction temperature drop, set to a fixed value; Tmat_brittle: Embrittlement temperature of hydrogen storage material; The temperature change rate threshold is determined based on the hydrogen storage stage, which includes a hydrogen absorption stage and a hydrogen release stage; when in the hydrogen absorption stage, the temperature change rate threshold is determined based on the ambient humidity.

3. The safety monitoring method for a solid-state hydrogen storage test bench according to claim 2, characterized in that, The Peq(Tenv) is obtained by querying the PCT curve or by calculating the Van'tHoff equation. The PCT curve is obtained based on historical data statistics; The formula for calculating the Van'tHoff equation is as follows: ； ΔH: Enthalpy change of hydrogen absorption in hydrogen storage materials; ΔS: Entropy change of hydrogen absorption in hydrogen storage material; R: gas constant (8.314 J / L) ); T: Ambient temperature (unit: K, T=Tenv+273.15).

4. The safety monitoring method for a solid-state hydrogen storage test bench according to claim 1, characterized in that, The method of using AI to predict the subsequent data trend corresponding to the current sensor data specifically includes: The target LSTM prediction model is used to predict the subsequent data trend corresponding to the current sensor data; the subsequent data trend is the target pressure-temperature curve within a preset future time.

5. The safety monitoring method for a solid-state hydrogen storage test bench according to claim 4, characterized in that, The method also includes the training process of the target LSTM prediction model, specifically including: Acquire historical data and uniformly divide the time series windows; The pressure, temperature, ambient temperature, and ambient humidity rate of change features are extracted from the historical data and normalized to obtain the source domain data. Construct an initial LSTM prediction model and output the initial pressure-temperature curve for a preset future time based on the source domain data; Based on the initial pressure-temperature curve within a preset future time and the actual pressure-temperature curve within a preset future time obtained from the source domain data, the parameters of the initial LSTM prediction model are adjusted to obtain the current LSTM prediction model. The source domain data is input into the current LSTM prediction model, and the current pressure-temperature curve is output within a preset future time. Determine the deviation between the current pressure-temperature curve and the actual pressure-temperature curve within a preset future time period; If the deviation does not meet the deviation condition, the parameters of the current LSTM prediction model are adjusted, and the steps of inputting the source domain data into the current LSTM prediction model and judging the deviation between the current pressure-temperature curve and the actual pressure-temperature curve within a preset future time are repeated until the deviation meets the deviation condition. Then, the current LSTM prediction model is output as the target LSTM prediction model.

6. The safety monitoring method for a solid-state hydrogen storage test bench according to claim 1, characterized in that, The use of AI to perform visual analysis on the current video stream specifically includes: Visual analysis of the current video stream is performed using target visual models; the target visual models include a target color recognition model, a target YOLOv5 model, and a target ResNet18 classification model.

7. The safety monitoring method for a solid-state hydrogen storage test bench according to any one of claims 1-6, characterized in that, The method further includes: Automatic polling extension interface; If a new sensor is detected to be connected to the expansion interface, an identification command is sent to the new sensor. Obtain the identity information sent by the newly added sensor; The identity information is compared with the built-in compatible library to complete automatic identification; Based on the identified communication protocol type, the corresponding parsing library is automatically loaded; Negotiate and lock the optimal communication parameters with the newly added sensor; The system parses the raw data format of the newly added sensor, performs normalization transformation and anomaly filtering, and outputs standardized data that matches the test bench. Based on the measurement parameter type of the newly added sensor, a safety threshold template or a user-matched threshold is automatically assigned.

8. A safety monitoring system for a solid-state hydrogen storage test bench, implemented using the safety monitoring method for a solid-state hydrogen storage test bench as described in any one of claims 1 to 7, characterized in that, The system includes: The information acquisition module is used to acquire real-time information, including real-time sensor data and real-time video streams. The information preprocessing module is used to perform time-series alignment processing on the real-time information to obtain the current information; the current information includes the current sensor data and the current video stream. The real-time monitoring module is used to determine whether the current sensor data exceeds a dynamic threshold, and to use AI to predict the subsequent data trend corresponding to the current sensor data and to perform visual analysis on the current video stream. The safety rule triggering module is used to issue an alarm when the current sensor data exceeds the dynamic threshold; to issue a warning when the predicted subsequent data trend does not meet the dynamic conditions; and to determine the leak location and close the solenoid valve associated with the leak location when the analysis results obtained from visual analysis indicate a hydrogen leak.

9. An electronic device, characterized in that, It includes a processor and a memory, wherein the memory stores at least one instruction or at least one program, the at least one instruction or the at least one program being loaded and executed by the processor to implement the safety monitoring method for a solid-state hydrogen storage test bench as described in any one of claims 1 to 7.

10. A computer-readable storage medium storing at least one instruction or at least one program, the at least one instruction or the at least one program being loaded and executed by a processor to implement the safety monitoring method for a solid-state hydrogen storage test bench as described in any one of claims 1 to 7.

Images